基于多源数据融合的电动汽车充电站选址方法

In view of the defects of the existing machine learning methods for the location of electric vehicle charging stations, such as the weak correlation of driving factor selection, the simple labeling strategy and lack of relevant basis in the case of insufficient sample labels, and the low location accuracy caused by the single selection of similarity discriminant function in the relevant algorithms. Methods：This study takes six districts such as Xuhui District and Jing\'an District of Shanghai as examples, combined with PFAHP and TOPSIS methods, A LogitBoost location selection model based on hybrid similarity automatic labeling and semi supervised learning is constructed, and four groups of schemes and eight experiments are designed to verify the superiority and applicability of this method. Results：The experimental results of the model show that the accuracy of this model is improved by 0.6% - 2.6% compared with the single similarity and the model using artificial labeling, and the recall rate, F1 and sorting accuracy are improved to a certain extent. Conclusions：The test shows that the method proposed in this paper can be better used to solve the location problem of electric vehicle charging station than other existing methods. At the same time, it also has a certain reference value for the location of other public facilities.